Abstract
Multiple-antibiotic resistance in the pneumococcus, defined as resistance to ⩾3 classes of antibiotics, first emerged in South Africa in 1978 [1]. The selection of these multiply resistant bacteria was nosocomial, and spread outside of the hospital was not demonstrated. It is a reasonable assumption that antibiotic-resistant strains have, at least initially, no compensatory mutations to give them an advantage in the absence of antibiotic selective pressure. After the initial antibiotic selection, they are, therefore, diluted in the population of susceptible strains. A single exposure to an effective antibiotic thus would be expected to kill susceptible strains and select preexisting resistant bacteria. Over time, however, the proportion of resistant strains would then be reduced as their carriage is terminated by natural immunity. Evidence of such a process is available from a study of selection of resistant pneumococci in an aboriginal population exposed to a single dose of azithromycin for the treatment of trachoma [2] and from a recent comparative study of the impact of amoxicillin-clavulanate versus that of azithromycin on the nasopharyngeal flora of children in Texas who had acute otitis media [3].